Japan Airlines tests humanoid robots for ground operations, will transport cargo on the tarmac at Haneda Airport

Japan Airlines (JAL) will launch a humanoid robot pilot test at Tokyo Haneda Airport in May, with Unitree G1 and UBTECH Walker E responsible for transporting cargo containers. The test is scheduled to continue until 2028; the long-term goal includes cabin cleaning and ground support device operations to address labor shortages caused by the surge in inbound travelers.
(Background summary: Robot industry visions: the integration and evolution of automation, artificial intelligence, and Web3)
(Additional context: I give up, AI robots have won)

Table of Contents

Toggle

• What is being tested, and how?
• Why choose humanoid robots instead of other types?
• How far is the commercialization of humanoid robots?

Two humanoid robots, approximately 1.3 meters tall and weighing 35 kilograms, are about to transport cargo containers on the tarmac at Tokyo Haneda Airport. This is part of Japan Airlines’ (JAL) planned pilot test starting officially in May 2026.

The driving force behind this is straightforward: the number of inbound travelers has surged in recent years, but the staffing gap for airport ground operations is becoming increasingly difficult to fill. JAL’s response is to turn to humanoid robots capable of standing and walking on two legs.

What is being tested, and how?

According to an official JAL press release, this pilot is jointly led by JAL’s subsidiary JAL Ground Service and GMO AI & Robotics Corporation, with testing scheduled from May 2026 to 2028.

The robots involved are two models: China’s Unitree Technology’s G1, and UBTECH’s Walker E. Specifications for G1 show a standing size of approximately 1,320 × 450 × 200mm, weighing about 35 kilograms—lighter than an adult but large enough to operate in typical work environments.

The initial focus of the test project is on cargo container transportation. JAL specifically states in the announcement that, at this stage, the robots will not directly handle passenger luggage; starting with cargo operations, which have a more fixed structure, is a pragmatic way to reduce testing risks.

The long-term vision is broader: operations such as cabin cleaning, luggage cart handling, and other ground support devices are also part of the plan. If the tests go well, it’s not impossible to see robots working on the tarmac at Haneda before 2028.

Why choose humanoid robots instead of other types?

Factories and warehouses already extensively use robots, but those are usually fixed robotic arms at specific workstations or autonomous guided vehicles (AGVs) moving along fixed paths. Their common features are predictable environments and repetitive tasks.

In contrast, airport ground operations are quite different. The tarmac has complex movement patterns, cargo varies in shape and weight, and unexpected situations frequently occur. Deploying traditional specialized robots would require significant modifications to the work environment, which involves high costs and time.

JAL’s approach is: Let robots adapt to human environments, rather than making environments fit robots. Humanoid robots are chosen because they are designed to move within human workspaces: climbing stairs, opening doors, operating tools designed for human hands—all within their capabilities (at least in theory).

This pilot also aims to verify one key point: whether humanoid robots equipped with the latest AI models can truly adapt to open, dynamic airport environments without extensive site modifications.

How far is the commercialization of humanoid robots?

Humanoid robots are not a new concept, but achieving stable operation in real-world work environments is another matter.

In recent years, robots like Tesla’s Optimus, Figure AI, and Boston Dynamics’ Atlas have appeared and been tested in automotive factories and warehouses. However, most public demonstrations still occur in controlled environments, and large-scale commercialization has not yet arrived.

The significance of JAL’s pilot is that airports are highly complex environments with real commercial pressures. Unlike warehouses, which can be fully enclosed, or factories with highly standardized processes, airports present unique challenges. If Unitree G1 and UBTECH Walker E can deliver usable results during the Haneda tests, it will provide a more convincing data point for the entire humanoid robot industry than factory pilot projects.

The testing will continue until 2028, and results will need time to validate.